Swarm Intelligent Algorithm In The Rainstorm Intensity Formula Research And Application

Rainstorm intensity formula is to reflect the law of rainfall and guide the city drains the important foundation of flood control drainage engineering design and related facilities construction,but with global warming,increasingly frequent natural disasters and climate changes,especially the waterlogging problems aren't improved in recent years.High precision of the rainstorm intensity formula not only provide reliable basis for the design of the urban drainage facilities,but also for urban rainwater utilization,rainstorm runoff control design has very important significance.This study in order to Baoji weather bureau of meteorology 31 years' self-recording rainfall data as the foundation,through to weather the bureau of meteorology selfrecording rainfall data to determine the best sample selection method.Then through three different frequency adjustment model and three different optimization algorithm to get the highest precision of rainstorm intensity formula.At the same time,according to the rainfall data and the latest derive rainstorm intensity formula of conform to the Baoji type of short duration design storm rain has important application value to predict the occurrence of heavy rain.In the process of research obtained conclusions as follows:(1)The lack of independence in multiple sample method,and annual maximum method is easy to ignore rainstorm in the heavy rain.At present,most of the meteorological and hydrological departments only have maximum and reached more than 20 years of rainfall data,and meet the conditions of maximum method.But method is lack of the required data in multiple samples,need to be a large number of collection,statistics big workload,time-consuming.Therefore,the study is recommended to use in the maximum method.(2)Through Pearson ? type frequency model,Index model and Gumbel frequency model of annual maximum method respectively selected to calculate sample the average absolute mean square error and the average relative mean square error can obtain: Pearson type ? frequency model SQ1=0.086 mm/min?SQ2=7.742%;Index frequency model SQ1=0.095 mm/min ? SQ2=8.447%;Gumbel frequency model SQ1=0.047 mm/min ?SQ2=5.327%.Clearly,the average absolute mean square error and the average relative mean square error from Gumbel frequency model is less than the rest of the two methods,and have a better imitative effect.Therefore,it get a highest precision data tables about return period-rainstorm intensity-rainfall duration(i-P-t).(3)Based on MATLAB development function formulation of appropriate application and combined with the i-P-t data tables,data import,method selection,parameter setting method and the optimization algorithm for instance with different return period of storm intensity formula of A?C?b?n.Error analysis shows: Particle swarm optimization(PSO)?=0.035 mm/min,?=3.815%;Genetic algorithm(GA)?=0.048 mm/min,?=5.187%;Traditional algorithm(TR)?=0.079 mm/min,?=8.507%.At the same time,to the short duration(2~10 a and 2~20 a)error analysis indicates that the particle swarm algorithm to derive the rainstorm intensity formula precision are higher than that of genetic algorithm and traditional algorithm.Ultimately derive rainstorm intensity formula of the unknown parameter A=5.998?C=2.496?b=13.531 and n=0.831.Thus,the final rainstorm intensity formula have been confirmed.(4)Through the comparison analysis to determine two heavy rain distribution type choice in Keifer&Chu rain derive Baoji storm rain distribution.Meanwhile,type and rain Pilgrim&Cordery method for operation and management of reference.Through the analysis on characters of short duration storm hyetograph is unimodal(peak in the first half),the rain peak position coefficient r=0.420,before and the peak is slow after the trend of the steep slope,slow after the peak after the first steep.